The most revealing thing about Diodes Incorporated is how rarely a consumer needs to know its name. A power switch wakes the right circuit. A clock generator keeps data arriving on beat. A Hall sensor notices that a motor moved. A ReDriver cleans up a signal before it turns into digital soup. These are supporting roles, and Diodes has collected an unusually large cast: more than 28,000 products sold to over 50,000 customers.
The Plano, Texas company made $1.482 billion in 2025 by supplying the semiconductors that sit between a system's glamorous brain and its messy physical life. Its parts regulate voltage, route data, protect components, sense position and translate one electrical language into another. They go into cars, factories, servers, telecom gear, appliances, wearables and chargers. If a machine has power, motion or a fast cable, Diodes probably has a catalog page for one of its headaches.
That sounds like a parts-bin business. It is also a portfolio strategy, a manufacturing strategy and an acquisition strategy folded together. The company does not need every new design to contain one expensive hero chip. It can win several modest sockets on the same circuit board, then use distribution and application engineers to repeat the trick at industrial scale.
01 / What it actually does
The company sells fewer reasons to call the hardware team at midnight
Start with power. Diodes sells rectifiers, MOSFETs, AC-DC and DC-DC converters, voltage regulators, LED drivers, load switches and protection devices. The job is mundane only until voltage spikes, heat rises or a battery drains while the product is asleep. Then efficiency, package size and protection circuitry become the whole meeting.
Move to signals. Its logic devices translate voltage levels. PCIe switches decide where high-speed data goes. ReDrivers repair degraded signals over boards and cables. Clock ICs and crystal oscillators keep servers and network equipment synchronized. Sensors measure current, temperature, position and magnetic fields. The products are individually narrow; together they cover much of the connective tissue of an electronic system.
not revenue share
Automotive shows why breadth matters. A modern vehicle needs chips for headlights, displays, USB-C charging, motor control, ADAS power rails and high-speed cockpit links. Qualification can take time and money, but once a supplier proves reliability and gets designed in, switching is not casual. Diodes can approach an automaker or Tier 1 supplier with a group of products rather than a lonely part number.
Industrial customers have a similar bias toward reliability and long availability. Computing buyers care about power density, timing and signal integrity. Consumer products care about size and cost. Communications gear cares about speed and robustness. Diodes changes the pitch by market, but the underlying move stays constant: solve more adjacent problems on the customer's board.
02 / The compounding acquisition
It bought capabilities the way a careful cook buys knives
Diodes began in 1959, but the modern company is a sequence of deliberate additions. The 2008 purchase of Zetex, valued at $176 million, strengthened discrete and analog technology while adding European reach. Power Analog Microelectronics arrived in 2012, followed by BCD Semiconductor in 2013, broadening high-voltage, linear and power-management ICs.
The 2015 Pericom deal was the hinge. For roughly $400 million, Diodes gained timing, switching and signal-integrity products. The catalog moved deeper into the fast digital connections running through computers and communications equipment. In 2020, the roughly $428 million Lite-On Semiconductor transaction added power discretes, manufacturing scale and more Asian operations. A Greenock, Scotland wafer fab and onsemi's South Portland, Maine facility expanded internal fabrication. Fortemedia, acquired for $52.6 million of net consideration in 2024, brought voice-processing technology aimed at cars and computers.
analog + discrete
timing + links
power + scale
voice processing
deal pending
The pattern is more interesting than the shopping total. Each deal adds something that can travel through the existing sales network: a product category, process technology, factory, geography or customer relationship. The acquisition does not have to invent a second company. It makes the current company more useful to the same engineer.
“My top priority remains on delivering accelerated growth, while expanding margins and profitability.”Gary Yu, on becoming CEO in May 2025
Yu is an unusually legible choice for that job. Before becoming chief executive, he ran business groups, Asia-Pacific sales, a Shanghai fab, the BCD unit and the integration of Lite-On Semiconductor. His résumé traces the machinery that Diodes is trying to coordinate: customer demand, product portfolio, acquired teams and factory loading.
03 / The business model
Distribution supplies reach; factories supply control
Diodes sells through its own teams, independent representatives and distributors. In 2025, distributors generated about 65 percent of sales; direct and electronics-manufacturing-services customers made up roughly 35 percent. That channel mix is how a company can serve more than 50,000 customers without placing an account executive beside every soldering iron.
Its difference is the hybrid manufacturing model. Diodes operates wafer fabs and assembly-and-test sites, but it also buys from external foundries and subcontractors. Internal plants can improve cost, process control, capacity assurance and packaging know-how. Outside capacity adds flexibility and technologies the company does not own. In a supply squeeze, having both is practical leverage.
The small-package expertise matters more than it photographs. Engineers want components that occupy less board space, shed heat and handle more power. Shrinking a package while preserving electrical and thermal performance can earn a design win even when the underlying function is familiar. In a catalog business, packaging is not wrapping. It is product development.
04 / What failed first
When the cycle cooled, the factory math got loud
The model has a clean weakness: factories keep costing money when orders slow. Semiconductor customers entered a broad inventory correction after the pandemic-era shortage. Diodes' revenue fell from $1.662 billion in 2023 to $1.311 billion in 2024. Lower utilization pressured gross margin because depreciation, people and facility costs could not shrink at the same speed as shipments.
What changed management's posture was not a philosophical conversion. It was evidence that channel inventory was improving and demand was returning. Revenue rose 13 percent to $1.482 billion in 2025. In the second quarter of 2026, sales reached $445.5 million, up more than 20 percent from a year earlier, while GAAP gross margin rose to 33.1 percent. Management guided the next quarter to roughly $510 million at the midpoint and 35 percent gross margin, explicitly linking the improvement to better factory utilization and product mix.
The operating-leverage bargain
Orders up: fixed factory costs spread across more units, helping margin. Orders down: the same costs land on fewer units, hurting margin. Internal manufacturing is an advantage only when capacity, demand and product mix remain in conversation.
The first crack, then, was not that customers stopped needing power or timing forever. It was loading. That distinction matters. A product problem demands invention; a utilization problem demands volume, mix, patience and cost control. Diodes responded with new automotive and computing products, tighter inventory and expense discipline, not an abandonment of manufacturing.
05 / The next aisle
Now it wants to sell chips to the machines that test chips
In July 2026, Diodes agreed to acquire ElevATE Semiconductor for $250 million in cash. ElevATE designs low-power, high-density ICs for automated test equipment. If the deal closes, expected in the second half of 2026 subject to approvals, Diodes says the business should add about $50 million of revenue in its first twelve months and grow more than 20 percent annually over the following four years.
The strategic logic is tidy. Automotive, data-center and AI chips are becoming more complex, so manufacturers need more capable test equipment. ElevATE sells semiconductor components into those testers. Diodes can add manufacturing scale, analog and power products, and a global sales network. It is the familiar playbook applied one level upstream.
Still, announced benefits are forecasts, not facts. Integration can lose people or customers. ATE demand can wobble with semiconductor capital spending. Regulators can delay closing. The deal works best if ElevATE's specialists remain focused while Diodes cross-sells without smothering the thing it bought.
What an operator can steal
- Bundle around the customer's workflow. Organize adjacent products around a real system problem, not internal departments.
- Buy a capability that travels. The best acquisition can move through an existing channel and raise revenue per relationship.
- Own the bottleneck selectively. Internal capacity helps when quality, supply or packaging creates an edge; outside capacity preserves flexibility.
- Measure utilization before celebrating scale. A factory, warehouse or service team is leverage in both directions.
- Make reliability a sales feature. Qualifications, support and consistent delivery can matter as much as a specification lead.
06 / Where the playbook breaks
Breadth is useful only if the pieces still fit
Diodes' approach would not work everywhere. It favors markets with many repeatable component needs, long product lives, demanding qualifications and customers who value a broad approved supplier. It is weaker where one fast-moving platform dominates, where manufacturing technology changes before acquired assets pay back, or where software-style margins cannot absorb factory overhead.
Catalog breadth can also become catalog clutter. Twenty-eight thousand products require documentation, inventory decisions, lifecycle management and sales training. Cross-selling fails when account teams cannot explain the portfolio or when acquired product lines share no customer. Vertical integration fails when a captive fab is less efficient than an external supplier and management keeps feeding it merely because it exists.
That is the sober version of Diodes' advantage. The company is not protected by invisibility, age or a large spreadsheet of part numbers. Its edge exists when those parts solve linked problems, reach customers cheaply, fit into qualified packages and keep factories productively loaded. In 2024, that system showed its cost. In 2025 and the first half of 2026, it showed its recovery mechanics.
The broad lesson is pleasantly unromantic. Important companies do not always make the object on the billboard. Sometimes they make the switch behind the display, the clock beside the processor and the protection device that prevents both from becoming smoke. Diodes has spent decades turning those small obligations into one large business.